Reviewer 3's Review

“The goal of the ms is very interesting: to look for regions in the genome outside the mtDNA and Y to work with the tools of phylogeography. Nonetheless there are major problems on it:

1.- One thing is that there is a very strong LD in a given regions and another, very different, is that not a single recombination event has taken place. Thus the very likely recombinations that have occurred in human history along the studied region are not seen, as there is no attempt to try to account for them. Thus a lot of problems may arise: the networks drawn and analyzed may have reticulations due to recombination. Any calculation coming from them would be biased.

It is note enough to say the "inspection of histograms reveals that our data are perfectly compatible with absolute LD"

If it is a non-recombining haplotype block, data from Hapmap should be shown and demonstrated (if possible) that no recombination has taken place, an extremely unlikely event.

2.- For coalescent analysis it is needed, before hand, to have THE true tree. I do not see how confident we may be that the tree assumed is THE real one (and the only one possible). With a few recombination events the whole analysis would change drastically.

3.- The way mutation rate is handled is not very precise. Differences among mutation rates may be very strong and have impact in the analysis, mainly if STRs with different number of nucleotides in the repeat unit are being used. The different values used in pg 12 is not clear where they come from of if they are just general values. To position the values to get a general mean decided beforehand is not accurated, mainly when handling a small number of STRs. Moreover the use of a stepwise mutation model is nice as a model but it is known that it is not true.

4.- Recombination rates are taken from the Decode map. No use is done of the HapMap map. Why that? May be even being in a region of relative low recombination rate there are a few recombination hot spots!

5.- Samples used from HGDP-CEPH should be referenced to the paper of Rosenberg in Annals of Human Genetics. Does not seem the case, and this is now totally necessary.

6.- In general keeping using trees to describe population history as a sequence of splitting events is not acceptable. It may be a working tool but not a way of understanding the dynamics of populations, as migration cannot be ignored. Thus the final tree of figure 4 is not clear the meaning it has: are these the population histories? Then bootstrap values would likely show that the tree s not robust. For example it is known that Central Asia is a mixture of East and West. To draw it with East may simply be misleading, and it would be better to show they relative position.

7.- Last part has a story-teller tone (pg 15-16), with expressions like: match perfectly, data in quite harmony.... These do not seem very serious results form the data, but a way of showing that results are similar to other previous results. Then, what are the news given by this study?”

N.B. These are the general comments made by the reviewer when reviewing this paper in light of which the manuscript was revised. Specific points addressed during revision of the paper are not shown.